closure on “a new protection scheme for fault detection,.pdf

7/27/2019 Closure on A New Protection Scheme for Fault Detection,.pdf

1/3

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003 653

(GPS) satellites. The authors should be commended for their paper in-vestigating the proposed scheme extensively by numerous simulationstudies.

The discusser would like the authors to further clarify and commenton the following points.

1) The proposed protective scheme includes considerable numberof components [e.g., phasor measurement unit (PMU), trans-mitter/receiver, GPS receiver, communication link, etc. which

could make the whole system complex]. Reliability of the pro-tective system might decrease due to its high complexity.

2) In the case that GPS signal is not available, how does the algo-rithm behave? What kind of backup is considered for protectingthe transmission line?

3) The communication link transmission delay is considered andcompensated in the algorithm. Is the algorithm able to compen-sate nonconstant communication link delays?

4) Have the authors compared this algorithm with other transmis-sion line pilot relaying algorithms? It is suggested that perfor-mance of the proposed algorithm would be compared to perfor-mance of the phase comparison scheme in a sequel work.

5) According to section III, Fig. 4, the algorithm has to perform aconsiderable amount of calculations in each sampling interval.

Different tasks such as fault detection, fault classification, modaltransformation, online parameter estimation, and direction dis-crimination should be performed in approximately 0.5 ms (32samples/cycle). Have the authors experienced which kind ofhardware should be used to implement the whole procedure?

6) Performance of the algorithm for high-impedance faults is eval-uated using a three phase to ground fault with 1-k

fault resis-tance. This kind of fault on transmission lines is not that usual.It would be better to consider a more usual fault (e.g., single-line-to-ground fault with some fault resistance).

7) The CT saturation effect is studied and the relay response is pre-sented in Fig. 9. According to the paper, the fault location isset at 0.95 p.u. of the protected line. For remote faults, currenttransformers usually do not get saturated if they are properly se-lected. According to Fig. 9, the ratio of the fault current to theload current is approximately ten. It might be better to test thealgorithm with more close-in faults to study CT saturation effect.

8) Have the authors tested the algorithm with any evolvingfault case (e.g., phase-to-ground a-g fault evolving to twophase-to-ground a-b-g fault in a few milliseconds)? How doesthe scheme behave during transmission line switching andenergization? These cases are worth to be further addressed infuture research.

Closure on A New Protection Scheme for Fault Detection,

Direction Discrimination, Classification, and Location in

Transmission Lines

Joe-Air Jiang, Ching-Shan Chen, and Chih-Wen Liu

We would like to thank the discusser for his interest in our paper 1

and for his valuable and constructive comments. We first offer com-ments of a general nature, and then respond to questions in the order

asked.

Real-time wide area measurements from phasor measurement units

(PMUs) allow for innovative solutions to traditional utility problems.

The synchronized phasors from substations are then collected at a

central location. By collecting these precise synchronized measure-

ments, techniques for monitoring, protection, and control of the power

system can be developed and implemented [1], [2]. For example, ABB

is presently developing PMU-based products and advanced solutions.

PMU functions are also being integrated into ABB protective relays

[1]. Better relaying and fault location performance can be achieved

using multiterminal synchronized measurements. Based on this trend,

a new PMU-based protection scheme is presented for transmission

line protection.

1) We agree with the discussers opinion that reliability is one of

the basic considerations for designing relays. Digital relays over-

come this problem by constantly executing self-monitoring rou-

tines in the background of protection processing. If a failure

occurs, an alarm is immediately issued, and a specific failure

message can be communicated to the maintenance personnel.

According to the descriptions about applications of global po-

sitioning system (GPS) to relaying and control in [2], we can

say that the reliability of the GPS system is nearly ideal. When

the GPS signal is interrupted or incorrect within a predefined

time, an alarm signal will be sent, and the synchronized sam-

pling method will be changed into a traditional version. It is

also important to detect channel failures and promptly activatebackup protection such as distance protection during such con-

ditions, and to return to normal when channel quality is restored.

The reliability of the proposed scheme can be further improved

by adding redundant communication channels. In particular, the

use of digital communications and numerical relays enabled us

to provide a greatly enhanced relaying scheme together with net-

work supervision and control.

2) When the protection scheme with GPS for data synchronization

is proposed,the likelihoodof theloss of GPS signals andantenna

failure should be considered. In practice, when the GPS antenna

is well sited, the loss of GPS signals is rare. Moreover, it is also

possible to run the traditional ping-pong technique for propaga-

tion delay measurement as a continuous background task even

when the GPS signal is lost [3]. Digital technology offers inclu-sion of many protection functions in a relay unit. When the GPS

signalis notavailable,it is still possible to maintain protection by

Manuscript received November 29, 2001.J.-A. Jiang is with the Department of Bio-Industrial Mechatronics Engi-

neering, National Taiwan University, Taipei, 106, Taiwan, R.O.C.C.-S. Chen and C.-W. Liu are with the Department of Electrical Engineering,

National Taiwan University, Taipei, 106, Taiwan, R.O.C.Digital Object Identifier 10.1109/TPWRD.2003.809882

1J.-A. Jiang, C.-S. Chien, and C.-W. Liu, IEEE Trans. Power Delivery, vol.17, pp. 34-42, Jan. 2003

0885-8977/03$17.00 2003 IEEE

Authorized licensed use limited to: Iran Univ of Science and Tech. Downloaded on June 30, 2009 at 09:38 from IEEE Xplore. Restrictions apply.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


2/3

654 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003

Fig. 1. Relay response for a phase-a to ground fault with a high faultresistance. (a) The behaviors of modal fault detection index j M j underphase-a basis. (b) The response curves of a b s [ a r g ( E = B ) ] . (c) and (d)are the behaviors of modal fault detection index under phase-b and -c basis,respectively.

using the nonunit elements, distance relays, or directional over-

current, as backup protection within the same relay and alterna-

tively enabled on communication channel failure. This diversity

of operationprinciples in thesameunit enhances theoverallrelay

performance without a significant cost increase.

3) Communication time delay produces a phase shift between the

local and remote signals in unit protection systems. One of the

widely used techniques to align the signals is known as the

pingpong technique, which measures the time delay between

the sampling pulses at two different locations. Once the delay

between the two sampling pulses is measured, the two signals

corresponding to these two sampling pulses are time aligned.

The ping-pong technique assumes that the forward time delayis equal to the return time delay [4]. However, the proposed

relaying scheme uses an external reference signal (i.e., GPS

signal, to synchronize the line measurements at both ends).

This method removes the need to measure and compensate for

the communication delay. Therefore, the effect of nonconstant

channel delay on the protection algorithm can be ignored.

4) The authors would like to point out that the work presently being

carried out is preliminary research. Extensive simulations under

various system and fault conditions are performed to verify the

performance and correctness of the proposed relaying algorithm.

The studies of the proposed scheme compared with other trans-

mission line pilot relaying algorithms are not considered here.

We agree with the discussers opinion that perhaps a future work

may undertake a comparative study.5) Since the proposed algorithm is performed in a sequential

method, the various tasks such as fault detection, direction

discrimination, and fault classification are not executed si-

multaneously in each sampling interval. Moreover, the online

parameter estimation is performed at a predetermined interval

before a fault occurs. The algorithm also can be implemented in

a lower sampling rate such as 16 or 12 samples per cycle but may

take more relaying response time. Now, the algorithm is only

programmed using MATLAB in a personal computer to check

its correctness. In the future, we will implement the algorithm

using C language or assembly language on a digital signal

processor (DSP) board. With the advancements of high-speed

DSPs, we believe that the algorithm can be real time executed.

Fig. 2. Relay response for an evolving fault. (a) The behaviors of modalfault detection index j M j under phase-a basis. (b) The response curves ofa b s [ a r g ( E = B ) ] . (c) and (d) are the behaviors of modal fault detectionindex under phase-b and -c basis, respectively.

6) We agree that a three-phase to ground fault with 1 k

fault

resistance on transmission lines is not usual. Following the

discussers comment, the authors show a relay response for a

phase-a to ground fault. The fault resistance is 300

, and

fault position is set at 0.5 (p.u.). As shown in Fig. 1(a) and (b),

the proposed relaying scheme correctly detects the fault and

discriminates the fault as an internal fault. From the patterns

shown in Fig. 1(a), Fig. 1(c), and Fig. 1(d), it is obvious that the

fault type is phase-a to ground fault.

7) The discusser seems to misunderstand the definition of fault lo-

cation used in this paper. As shown in Fig. 5 in this paper, the

power flowdirection isfrombus A to bus B and the fault position

of 0 p.u. is defined at receiving end (i.e., bus B). For more detaildescriptions, the discusser can refer to our previous research [9].

Fig. 9 shows a relay response for a fault occurring at 0.95 p.u.

of the protected line, and the fault does not belong to a remote

fault. Moreover, the ratio of the fault current to the load current

is about 15 rather than 10. We agree that current transformers

(CTs) will not get saturated if they are properly selected. In this

paper, we purposely make CTs saturation to test performance of

the algorithm under this situation.

8) It is very important that the faulted-phase selector is capable of

changing the decision in the event of an evolving fault. Fig. 2

depicts an evolvingfaultresponsein which a phase to ground ag

fault subsequently evolves into a twophaseto ground abg fault

approximately 10 ms after initial fault inception. From Fig. 2, we

observe that thescheme will fast detect thefault, discriminate thefault as an internal fault, and correctly identify the faulted phase

(a-phase) after the ag fault occurrence. After time of 10 ms, the

beta-mode signal in Fig. 2(a) is also greater than the predefined

threshold. At this time, the fault type will be identified as two

phase to ground fault and then a three-phase tripping command

is issued. Thus, the scheme can work well in case of evolving

faults.

Reclosing a faulted line or transmission line energization would re-

sult in a high inrush current. These cases should be recognized by the

relay. Through preliminary simulation studies, if a faulted line is re-

closed, the proposed scheme can correctly detect the fault again. When

a transmission line is energized, the value of the fault detection index

will be greater than thepredefined threshold and then decays with time.

http://-/?-http://-/?-


3/3

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003 655

Under this condition, the behavior of the scheme is similar to an ex-

ternal fault case. Utilizing the fault detector incorporated with appro-

priate threshold setting and the fault directiondiscriminator, the scheme

does not issue a tripping command. Indeed, these cases need to be fur-

ther investigated and are now under progress through extensive simula-

tions.We hope to reportthe results in thefuture. In addition,a technique

to discriminatebetween the line charging inrush current and thecurrent

resulting on closing or autoreclosing a faulted line is proposed in [5].

We also can use the technique to supplement our proposed relaying

scheme. If the case is recognized as closing a faulted transmission line,

the breaker should be tripped again. However, if the case is identified

as closing on line energization, the relay does not trip the breaker and

then our scheme takes over its normal function.

Once again, we would like to thank the discusser very much for his

insightful and valuable comments on our paper.

REFERENCES

[1] D. G. Hart et al., PMUsA new approach to power network moni-toring, in ABB Rev., 2001, pp. 5861.

[2] Synchronized sampling and phasor measurements for relaying andcontrol, IEEE Trans. Power Delivery, vol. 9, pp. 442452, Jan. 1994.

[3] S. H. Richards, S. Potts, N. Robinson, S. Pickering, and A. Apostolov,

Application Benefits of GPS Synchronized Current Differential Protec-tion, UK: ALSTOM T&D Protection and Control Ltd., 2000.

[4] H. Y. Li et al., A new type of differential feeder protection relay usingthe global positioning system for data synchronization, IEEE Trans.Power Delivery, vol. 12, pp. 10901097, July 1997.

[5] M. M. Eissa and O. P. Malik, Experimental results of a supplemen-tary technique for auto-reclosing EHV/UHV transmission lines, IEEETrans. Power Delivery, vol. 17, pp. 702707, July 2002.

Discussion of Frequency Characteristics of Wavelets

Chen Xiangxun

In the above paper,1 wavelet transform (WT) is to measure the local

similarity between waveforms of the signal and the wavelet [1], [2]. It

is a misunderstanding that any wavelet is suitable for any signal and

any applications. Choosing or designing the right wavelet is crucial for

a successful WT application. This is the problem that many authors

in power community have not paid attention to, but the author of the

paper and a few others [3], [4] did pay attention to. However, the cri-

terion for choosing the right wavelet depends on the application. For

example, compressing data and extracting feature by WT require dif-

ferent wavelets. Not one criterion including the number of coefficients

or the bandwidth characteristics of a wavelet suggested in this paper is

suitable for all applications.The number of coefficients and the bandwidth characteristics at

different scales of a wavlet are not certainly related. There are many

wavelets, which correspond to various bandwidth characteristics

but with the same number of coefficients. Furthermore, frequency

Manuscript received June 4, 2002.The author is with The National Engineering Research Center for T & D,

China Electric Power Research Institute, China.Digital Object Identifier 10.1109/TPWRD.2003.809893

1C. Parameswariah and M. Cox, IEEE Trans. Power Delivery, vol. 17, pp.794-799, July 2002

characteristic of a wavelet includes magnitude spectrum (MSP) and

phase spectrum (PSP). A variety of wavelets has the same MSP but

different PSPs. It can be proved [5] that the number of the effective

WT coefficeints (i.e., the energy distribution at WT-domain depends

not only on the MSP but also more strongly on PSP). The bandwidth

characteristic only involves the MSP. Obviously, it is not complete to

characterize the frequency characteristic of a wavelet by its MSP.

Even for MSP, it is also not complete to characterize MSP by band-

width characteristic. For a given bandwidth, we have a variety of MSP

with different decay toward high frequency (DHF) and different decay

toward low frequency (DLF). The DHF corresponds to the regularity

of the wavelet and the DLF to the number of vanishing moments [6].

Both of them are important for various applications of wavelets.

As for minimizing the energy leakage at WT-domain, we have two

choices. One is to make the energy concentrate on a few scales. The

other is to make the energy concentrate on a few coefficients at given

scales [7]. The two choices correspond to different requirements on

the wavelet. It seems that choosing a wavelet from the point of view of

bandwidth may not be the best way.

REFERENCES

[1] M. Unser and A. Aldroubi, A review of wavelets in biomedical appli-cations, Proc. IEEE, vol. 84, pp. 626638, 1996.

[2] O. Chaari et al., Wavelets: A new tool for the resonant grounded powerdistribution systems relaying, IEEE Trans. Power Delivery, vol. 11, pp.13011308, July 1996.

[3] A. Galli and G. Heydt, Discussion of wavelet and electromagneticpower system transients, IEEE Trans. Power Delivery, vol. 11, pp.10571057, Apr. 1996.

[4] C. Xiangxun, Wavelet-based detection, localization, quantification andclassification of short duration power quality disturbances, in 2002IEEE Power Eng. Soc. Winter Meeting, New York, NY, Jan. 2731,2002.

[5] , Complex compactly supported orthogonal wavelets and their ap-plications in power systems (in Chinese), Proc. Chinese Soc. Elect.

Eng., vol. 20, no. 7, pp. 8388, 2000.[6] W. Sweldens, Wavelets: What next?, Proc. IEEE, vol. 84, no. 4, pp.

680685, 1996.

[7] I. Daubechies, Ten Lectures on Wavelets Philadelphia, PA, 1992.

Closure on Frequency Characteristics of Wavelets

Chethan Parameswariah and Mickey Cox

We thank Dr. Xiangxun for his reply and interest in our paper1 Fre-

quency Characteristics of Wavelets. We agree that a proper choice or

design of wavelet depends on the application; a specific wavelet may

be an excellent candidate for one application and be a poor choice fora different application. In our opinion, this issue has not received suf-

ficient attention and this is what we intended to understand and clarify

in our paper by looking at one particular propertythe coefficients of

wavelets and the energy distribution.

Manuscript received August 25, 2000.The authors are with the Department of Electrical Engineering, Louisiana

Tech University, Ruston, LA 71272 USA.Digital Object Identifier 10.1109/TPWRD.2003.809744

1C. Parameswariah and M. Cox, IEEE Trans. Power Delivery, vol. 17, pp.794-799, July 2002

0885-8977/03$17.00 2003 IEEE

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-

closure on “a new protection scheme for fault detection,.pdf

Documents